Several major transitions in shell pattern and morphology can be observed during the life of Haliotis asinina. The initial differentiation of biomineralising cells is likely to include a localised thickening of the dorsal ectoderm followed by an invagination of cells to form the shell gland 32. The shell gland then evaginates to form the shell field (Fig. 1A) which expands through mitotic divisions to direct the precipitation of calcium carbonate (CaCO₃) via the secretion of organic molecules. In this way the larval shell (protoconch) is formed (Fig. 1B and 1C). The construction of the haliotid protoconch is complete following torsion (Fig. 1C), and remains developmentally inert until the animal contacts a specific cue that initiates the process of metamorphosis. 3334. The transition from protoconch to teleoconch (juvenile/adult shell) is clearly visible at metamorphosis (Fig. 1D), and suggests the action of a different biomineralising secretome. The early postlarval shell is more robust and opaque than the larval shell but has no pigmentation. Juvenile H. asinina begin to develop a complex colouration in the shell several weeks after metamorphosis (Fig. 1E and 1F). This pattern is gradually lost with growth as the shell becomes thicker and more elongate (Fig. 1G and 1H). These large scale morphological changes are accompanied by mineralogical and crystallographic changes (Fig. 1I–L ). Well defined tablets of nacre are present in shells larger than approximately 5 mm (Fig. 1J–L) which are absent or poorly resolved in shells 1 mm or less (Fig. 1I). In larger shells, a ventral cap of CaCO₃ that underlies the tablets of aragonitic nacre continues to thicken (Fig. 1K and 1L).

We have isolated nine genes that are expressed in cells at the anterior edge of the mantle in the region responsible for the construction of the tropical abalone shell (see Methods and 28). The temporal expression profiles of some of these genes coincide with ecological transitions that occur during development (Fig. 2). Has-ubfm, Has-calmbp1, and Has-cam1 transcripts are maternally provided to the egg, and appear to be constitutively expressed during development. Has-ubfm, Has-ferrt, Has-calmbp1 and Has-cam1 are present during embryogenesis, with detectable levels of transcripts present from the trochophore stage onwards and in all mantle biopsies surveyed. In contrast, Has-tsfgr1, Has-vm1, Has-vm2, Has-lustA and Has-Som are transiently expressed during the life of H. asinina. Has-tsfgr1 is highly expressed in larval stages followed by a down-regulation in 4 mm animals and a subsequent lack of expression in larger animals. Has-vm1 is highly expressed in veligers and juveniles ≦ 4 mm, but is absent in the mantle tissue of animals larger than 20 mm. Has-vm2 displays a strong up-regulation from trochophore to veliger stages, followed by a subsequent down-regulation in 20 mm animals. A smaller RT-PCR product is detected in 40 and 105 mm animals, suggesting the presence of an alternatively spliced transcript. Has-lustA and Has-Som transcripts are only detected in the mantle tissue of animals larger than 4 mm.

H. asinina has a pelagobenthic life cycle that includes a minimal period of three to four days in the plankton 3341. The first biomineralisation events occur shortly after hatching, with the fabrication of the larval shell (protoconch) over about a 10 h period. These structures allow the veliger larva to completely retract into a protective environment and rapidly fall out of the water column. The next phase of biomineralisation does not commence until the competent veliger larva contacts an environmental cue that induces metamorphosis 41. Postlarval shell (teleoconch) is laid down rapidly following metamorphosis with marked variation in the rate of its production between individuals. While the initial teloconch is not pigmented (Fig. 1D), it is textured and opaque such that postlarval shell growth is easily discerned from the larval shell (Fig. 1D inset). Subsequently, the teloconch rapidly develops a uniform maroon colouration similar to the crustose coralline algae (CCA) that the larva has settled upon (Fig. 1E). At about 1 mm in size further changes in the morphogenetic program of the mantle are reflected in the shell. Structurally, a pronounced series of ridges and valleys and a line of respiratory pores (tremata) have appeared (Fig. 1F). Furthermore, it is at this stage of development that the first recognisable tablets of nacre can be detected (Fig. 1J). Colourmetrically, the uniform maroon background is now interrupted by oscillations of a pale cream colour, and is punctuated by a pattern of dots (that only occur on ridges) which are blue when overlying a maroon field and orange when overlying a cream field. This shell pattern may enhance the juvenile's ability to camouflage on the heterogeneous background of the CCA they inhabit at this stage of development.

At 10 to 15 mm, this ornate colouration pattern begins to fade, with maroon and cream fields apparently blending to give a brown background. Blue and orange dots however persist on the ridges (Fig. 1G). With further growth, the ridge-valley structure fades to give rise to a smooth adult shell, with irregular brown-green triangles on a light brown background (Fig. 1H). These larger animals are nocturnal, graze amongst turf algae 42 and inhabit the undersides of boulders and coral bommies 43. Overall, ontogenetic changes in H. asinina shell pigmentation and structure match changes in the habitats occupied during development.

The spatial and temporal expression patterns of the nine genes investigated here reveal a complexity to the genetic networks that coordinate the deposition of larval, juvenile and adult shell. Many of the genes analysed in this study are expressed in the mantle during the production of larval, juvenile and adult shells (Has-Ubfm, Has-ferrt, Has-calmbp1), while others are restricted to one or two shell phases (Has-tsfgr1, Has-cam1, Has-vm1, Has-vm2, Has-lustA, Has-Som). While the lack of a detailed cell fate map through metamorphosis prevents conclusions from being drawn regarding cellular developmental homologies, the continuous expression of Has-tsfgr1 and Has-vm1 in cells no other than shell forming cells in both larval and postlarval stages suggests that a proportion of the postlarval mantle is derived from cells of the larval shell field.

Analyses of the expression profiles of the genes included in this study provide insight into the morphogenetic activity of shell production at different stages in the life of H. asinina. Genes that are continuously expressed in the mantle – Has-ferrt, Has-ubfm and Has-calmbp1 – are likely to play fundamental roles in biomineralisation. Has-ubfm encodes a highly conserved ubiquitin fold-like modifying protein 35 and is expressed in the expanding shell field suggests that specific intracellular processing of gene products is required to generate functional extracellular components of the biomineralising secretome. Two other evolutionarily conserved proteins, Has-ferrt and Has-calmbp1, are also expressed within the trochophore shell field and later in the mantle, and also are likely to be involved in intracellular events necessary for shell deposition. Iron is known to affect calcification processes in mammals 4445, algae 46 and molluscs 13. The high expression level of Has-ferrt in a range of cell types in the juvenile mantle is compatible with iron being essential for shell construction or pigmentation. Has-calmbp1 is similar to calcium dependent protein kinases, however its role in shell production remains unknown.

In contrast to these constitutively expressed genes, Has-tsfgr1 displays a dynamic expression profile in the shell forming tissue during development. The putative protein is composed of a set of glycine-rich repeats (over 52% Gly in the mature protein), suggesting it may possess elastomeric properties known to be important in various calcification processes 47. The highly repetitive nature of Has-tsfgr1 also suggests that this protein may be involved in forming the organic template upon which initial CaCO₃ nucleation occurs 48. Recently Yano et al. 14 isolated a family of glycine rich, repetitive motif proteins (Shematrins) from a mantle cDNA library of the pearl oyster Pinctada fucata. The Shematrin family is currently known to encode 7 proteins with similar C-terminal motifs which terminate in a tyrosine residue and are expressed in the mantle edge, apparently localised to the prismatic layer of the mature oyster shell 14. Interestingly, Has-tsfgr1 also possesses a C-terminal motif of 5 residues terminating in a tyrosine residue, suggesting that this feature may be of functional importance to this class of protein. Although sequence alignments of the Shematrins and Has-tsfgr1 do not reveal any close sequence homology, the high glycine content, repetitive nature and shared spatial expression suggest these proteins may play similar functional roles. Unlike Has-ubfm, Has-calmbp1, and Has-ferrt, which all maintain expression within juvenile and adult mantle tissue, Has-tsfgr1 is significantly down-regulated in the mantle tissue of >20 mm animals. This observation is compatible with different stages of shell development requiring the secretion of discrete sets of structural proteins, which act to alter the physical and mineralogical characteristics of the shell.

Two other novel genes – Has-vm1 and -vm2 – also display a dynamic temporal expression during the development of the shell. As expression of Has-vm1 is activated in the larval mantle after completion of the construction of the larval shell, this gene is likely to be involved in the construction of the postlarval shell following metamorphosis. This pattern of expression reveals a linkage between larval and postlarval mantles and is similar to that observed for the developmental regulator Has-Hox4 30. It also demonstrates that although larval shell synthesis has ceased, transcriptional activity in the larval mantle continues in anticipation for the next life cycle phase 49.

Has-vm2 is also differentially regulated during shell growth, and encodes a protein with the hallmarks of being involved in biomineralisation including a signal sequence, repetitive proline rich motifs and two putative O-linked glycosylation sites. Has-vm2 is also down-regulated in larger individuals, with a concomitant reduction in transcript size suggesting that alternative splicing of this gene product takes place in animals larger than 40 mm, again highlighting the different requirements for shell construction at different life cycle stages.

Three genes – Has-cam, Has-lustA and Has-Som – are expressed in patterns that are indicative of roles in biomineralisation during post-larval shell growth. Reflective of the various roles it plays within well studied mammalian systems including signal transduction and regulation of the cell cycle 50515253, Has-cam1 appears to play diverse roles during development. Highly expressed in the prototroch of trochophores, it is not until the veliger larva attains competence to metamorphose that Has-cam1 is detected within the mantle. This suggests that Has-cam1 is not directly involved in larval shell synthesis. In agreement with studies on bivalves 1254Has-cam1 is expressed within the gills and the outer fold of the mantle of juvenile animals. The expression pattern of Has-lustA supports its proposed role of binding aragonitic tablets of nacre together 1855, and coincides with the appearance of ordered aragonitic tablets. Interestingly, Has-lustA is down-regulated in the mantle tissue of mature abalone of 100 mm (Fig. 2) possibly reflecting a cessation of shell growth as this is close to the maximum size of 11 cm reported for this species 42. Has-Som has previously been shown to play a role in pigmentation of the juvenile shell 31 and is expressed in the mantle tissue of juvenile animals at the time complex colour patterning commences.

Many planktonic molluscan larvae face similar challenges during larval life and the evolution of a larval shell has clearly been a successful response to these challenges 56. On the molecular level, it is currently unknown the degree to which construction of the molluscan protoconch is conserved. Previous studies have revealed that the expression of the engrailed transcription factor in polyplacophoran 28, gastropod 272957 and scaphopod 26 representatives is restricted to cells that form boundaries between shell forming and non-shell forming ectoderm, suggesting that the regulatory mechanisms that establish shell forming structures in these clades were inherited from a common ancestor. Following metamorphosis, planktonic molluscan larvae inhabit a broad diversity of benthic ecological niches from sediments, coral reefs, deep sea hydrothermal vents and temperate rocky reefs. The shell of the tropical abalone undergoes several major transitions in morphology, mineralogy and pigmentation during its construction, each of which is adapted to suit the different habitats that larval, juvenile and adult forms occupy. These varied morphologies are the result of differential gene expression of both evolutionarily ancient and novel genes within the mantle tissue.

The genes investigated in this study were originally identified either through (1) an expressed sequence tag (EST) screen of genes expressed in the mantle of juvenile abalone 31, (2) an EST survey of developmentally expressed genes 58 or (3) a differential display analysis of developmentally regulated genes 33. Full length cDNA sequences for the genes used in this study were obtained using a RACE approach as described in Jackson et al. 33. cDNA sequences were initially characterised as described in Jackson et al. 31 and classified as either having conceptually derived amino acid sequence similarity with proteins in public databases, or encoding a novel secreted protein. The presence of signal peptides was inferred using the SignalP 3.0 server 59 and glycosylation predictions were made using the NetOGlyc server 60. Putative open reading frames (ORFs) for the evolutionarily novel and divergent genes Has-vm1 (

H. as

v

m

1

For genes encoding conserved proteins (Has-ubfm, ubiquitin fold modifier; Has-cam1, calmodulin; Has-ferrt, ferritin; Has-calmbp1, calcium binding protein; Has-lustA, Lustrin A), tBLASTx and BLASTp searches were conducted against the GenBank database using default settings (Has-Som, Sometsuke, has been previously characterised 31). Publicly available protein sequences that displayed significant similarity to H. asinina sequences and representing a broad taxonomic range were downloaded and aligned in ClustalX. Alignments were manually edited in MacClade. Percent identity and biochemical similarity for each sequence relative to the respective H. asinina sequence were calculated using the NCBI bl2seq algorithm 62.

Animals were procured from natural spawnings conducted at the Bribie Island Aquaculture Research Centre, Queensland, Australia as described in Jackson et al. 33. Larvae and juvenile abalone were relaxed in approximately 0.3 M MgCl₂ in FSW prior to fixation in 4% paraformaldehyde in 0.1 M 3-(N-morpholino) propane sulfonic acid pH 7.5 (MOPS), 2 mM MgSO₄, 1 mM ethyleneglycoltetraacetic acid (EGTA) and 0.5 M sodium chloride (NaCl) for 30 min at room temperature. Fixed samples were then rinsed several times with PBS buffer plus 0.1% Tween 20 and stepped into 75% ethanol and stored at -20°C. Decalcification of the juvenile shell was achieved by incubation in a solution of 4% paraformaldehyde, 1× PBS buffer and 350 mM ethylenediaminetetraacetic acid (EDTA) for 1 – 3 h depending on shell size. The remaining periostracum and proteinaceous components of the shell were manually dissected away from the animal with fine dissecting forceps. Whole mount in situ hybridisation using digoxigenin-labeled riboprobes synthesised from PCR templates was performed following Giusti et al. 63 and Jackson et al. 31.

Total RNA was extracted from eggs, 10 h old newly hatched trochophores, 134 h old competent veligers, whole 4 mm (shell length) juveniles, 20 mm juvenile mantle tissue, 40 mm juvenile mantle tissue and 100 mm adult mantle tissue using TriReagent following the manufactures instructions. cDNA was synthesised from 1 μg of intact total RNA following Jackson et al. 33. Relative levels of gene expression across the 7 cDNA samples were assessed by empirically determining the linear phase of PCR amplification for each gene using gene specific primers (available upon request). Briefly, each PCR was run for 20 cycles after which time 4 μl was removed. Reactions were then allowed to continue for a further 3 cycles and the process repeated until aliquots had been obtained from cycles 20 – 34. Samples were then separated on 2% agarose gels 64. Each PCR reaction was run in duplicate with cDNA synthesised in the absence of MMLV-RT as a control for genomic DNA contamination. Histone H1 was used as a constitutively expressed housekeeping gene and as an indicator of equivalent cDNA synthesis efficiency and PCR template quality 3365.

Nine, 10 and 11 h old trochophores were fixed in 3% glutaraldehyde in 0.1 M sodium cacodylate buffer for 30 min then washed in the same buffer prior to postfixing in 1% osmium tetroxide (OsO₄) in 0.1 M sodium cacodylate buffer. Samples were dehydrated through a graded series of ethanol before being infiltrated and dried overnight in hexamethyldisilasane (HMDS). The soft tissue of competent veligers and newly metamorphosed post-larvae was dissolved using 2.8% v/v sodium hypochlorite for approximately 5 min. The remaining shells were then washed extensively with de-ionised water and dehydrated with 100% ethanol before mounting. All samples were mounted either on double-sided tape or Leit-C conductive carbon cement on aluminium stubs and sputter-coated with gold. Samples were viewed with an S-2300 Hitachi scanning electron microscope at 10 kV.